Poses and Framing Release: One Detail Shot for Every Garment, and Poses from a Photo
Say you have everything set up in Brandmachine: your front view, your back view, your detail shot. You generate the detail shot. Now hold on. How should the AI model know what you want here?
The framing of a detail shot is completely different for jeans than for a shirt. On the shirt you might or might not want to see the head. Or you want 4% space between the head and the top edge. So you're back to prompting, and spending hours micromanaging your images.
What if it just worked? That's what we built.
Product Studio: one Detail shot for every garment
A Detail perspective now holds a framing per kind of garment: head to waist for a shirt, the waist down for a skirt. Brandmachine reads the garment type from the photos on upload, and almost all earlier products have one. Before you render, the board shows each look's framing, and a wrong type is one click to fix.
On the Breuninger Studio and Studio Sport sets there is nothing to set up: tops, outerwear and dresses get head to waist, everything else a close detail with no face. For now the top of the head is cut, because image models do not yet leave room above it.
Campaigns: show the pose instead of describing it
Give Brandmachine a photo of a pose, and the person in your look takes that pose. Described in words, "hands in pockets, weight on the left leg" comes out close, rarely exact.
The photo is redrawn as a plain mannequin first, so the bag or coat in the reference stays out. In our tests on GPT Image, the pose held with nothing extra carried over in 16 of 18 renders from the mannequin, against 16 of 36 from the photo.
Mark a moodboard image as Pose, and the agent puts it on the looks. Poses work on Campaign looks for now.